Responses API
Responses API is a server-side API product from OpenAI designed for building and managing AI model conversational interaction workflows. Key features include server-side maintenance of conversation history to reduce client-side context management overhead, support for integrated tool calls and file retrieval capabilities, and a unified interface for invoking multiple OpenAI language models. It is intended for developers building complex AI applications.
Core Facts
xAI provides an OpenAI-compatible API interface, which was the basis for early @ai-sdk/xai being able to adapt via the Chat Completions endpoint.
OpenAI于2024年推出了Responses API，引入了内置工具调用和状态管理能力
The OpenAI API supports Server-Sent Events streaming output with stream: true.
Timeline (last 90 days)
Mistral launched the Responses API in public beta, returning the model's answer, reasoning process and tool calls as separate entries in each response, supporting all chat models hosted by Mistral, including Large4.
GPT-6.1 Terra is compatible with the AI SDK, the OpenAI-compatible Chat Completions API, the Responses API, and coding agents connected to AI Gateway.
Developers can call Kimi K3.1 Pro via the AI SDK, Chat Completions API, Responses API, Anthropic Messages API, or a coding Agent connected to the AI Gateway.
The Responses API currently provides a cache hit guarantee within 30 minutes, and has launched a preview of a 12-hour cache window for some users.
The article suggests: if a project is calling the Responses API via Vertex AI, using Realtime WebSocket and encountering issues such as lost close state after connection recovery or connections not being properly released after cancellation, or still relying on Pydantic v1 for request validation, it should upgrade to v3.36.0 as soon as possible.
v3.36.0 adds Vertex AI support to the Responses API (PR #4139).
In Codex CLI 0.162.1, `wire_api = "chat"` causes a startup error because the new version removed the chat option and supports only the Responses API.
The default communication interface of the OpenAI Agents SDK is the Responses API, not Chat Completions.
Sending a Responses request to a gateway that only supports Chat Completions returns a NotFoundError (HTTP 404).
The OpenAI Responses API triggers a response.incomplete event when a response ends early (such as hitting the max_output_tokens limit), and that event also carries response.usage; listening only for response.completed will miss the usage of truncated answers.
36 more timeline events
All Facts (20)
xAI provides an OpenAI-compatible API interface, which was the basis for early @ai-sdk/xai being able to adapt via the Chat Completions endpoint.
90%VerifiedOpenAI于2024年推出了Responses API，引入了内置工具调用和状态管理能力
80%VerifiedThe OpenAI API supports Server-Sent Events streaming output with stream: true.
65%UnverifiedResponses API引入了previous_response_id参数实现多轮对话的自动上下文串联
85%UnverifiedThe Responses API is an evolved version of the Chat Completions API, integrating built-in tools such as web search, file search, and a code interpreter.
70%UnverifiedThe Responses API natively supports a variety of built-in tools, including web search, file search, code interpreter, and computer use.
60%UnverifiedThe Responses API is a next-generation API interface launched by OpenAI in 2025, intended to replace the Chat Completions API.
60%UnverifiedGPT-6.1 Terra is compatible with the AI SDK, the OpenAI-compatible Chat Completions API, the Responses API, and coding agents connected to AI Gateway.
50%UnverifiedDevelopers can call Kimi K3.1 Pro via the AI SDK, Chat Completions API, Responses API, Anthropic Messages API, or a coding Agent connected to the AI Gateway.
50%UnverifiedThe Responses API currently provides a cache hit guarantee within 30 minutes, and has launched a preview of a 12-hour cache window for some users.
50%UnverifiedThe article suggests: if a project is calling the Responses API via Vertex AI, using Realtime WebSocket and encountering issues such as lost close state after connection recovery or connections not being properly released after cancellation, or still relying on Pydantic v1 for request validation, it should upgrade to v3.36.0 as soon as possible.
50%Unverifiedv3.36.0 adds Vertex AI support to the Responses API (PR #4139).
50%UnverifiedIn Codex CLI 0.162.1, `wire_api = "chat"` causes a startup error because the new version removed the chat option and supports only the Responses API.
50%UnverifiedSending a Responses request to a gateway that only supports Chat Completions returns a NotFoundError (HTTP 404).
50%UnverifiedThe default communication interface of the OpenAI Agents SDK is the Responses API, not Chat Completions.
50%UnverifiedThe OpenAI Responses API triggers a response.incomplete event when a response ends early (such as hitting the max_output_tokens limit), and that event also carries response.usage; listening only for response.completed will miss the usage of truncated answers.
50%UnverifiedSwitching the same OpenAI Agents SDK Agent from the default Responses API to Chat Completions reduced format-error answers from 41 to 9, and switching to a public grounding system prompt reduced them to 20, indicating that API defaults are the main cause of format drift.
50%UnverifiedAmong the three endpoints of the Muse Wonder API, the reasoning effort parameter is named differently: reasoning.effort in Responses, and top-level reasoning_effort in Chat Completions.
50%UnverifiedThe Responses API provides two modes for maintaining state across turns: stateless encrypted replay (sending store: false and include: [reasoning.encrypted_content]) and server-side managed history (passing previous_response_id), and the two cannot be mixed.
50%UnverifiedThe Responses API is a new-generation interface introduced by OpenAI in 2025 to replace Chat Completions.
50%